项目：构建你的第一个向量搜索系统

from qdrant_client import QdrantClient, models
import os

client = QdrantClient(url=os.getenv("QDRANT_URL"), api_key=os.getenv("QDRANT_API_KEY"))

# For Colab:
# from google.colab import userdata
# client = QdrantClient(url=userdata.get("QDRANT_URL"), api_key=userdata.get("QDRANT_API_KEY"))

collection_name = "day0_first_system"
client.create_collection(
    collection_name=collection_name,
    vectors_config=models.VectorParams(size=4, distance=models.Distance.COSINE),
)

# Create payload index right after creating the collection and before uploading any data to enable filtering.
# If you add it later, HNSW won't rebuild automatically—bump ef_construct (e.g., 100→101) to trigger a safe rebuild.
client.create_payload_index(
    collection_name=collection_name,
    field_name="category",
    field_schema=models.PayloadSchemaType.KEYWORD,
)

points=[
    models.PointStruct(
        id=1,
        vector=[0.9, 0.1, 0.1, 0.8], # High affordability, high innovation
        payload={"name": "Budget Smartphone", "category": "electronics", "price": 299},
    ),
    models.PointStruct(
        id=2,
        vector=[0.2, 0.9, 0.8, 0.5], # High quality, high popularity
        payload={"name": "Bestselling Novel", "category": "books", "price": 19},
    ),
    models.PointStruct(
        id=3,
        vector=[0.8, 0.3, 0.2, 0.9], # High affordability, high innovation (similar to ID 1)
        payload={"name": "Smart Home Hub", "category": "electronics", "price": 89},
    ),
    # Add 2-5 more points to experiment with...
]

client.upsert(collection_name=collection_name, points=points)

# Define a query vector for "affordable and innovative"
query_vector = [0.85, 0.2, 0.1, 0.9]

# 1. Basic similarity search
basic_results = client.query_points(collection_name, query=query_vector)

# 2. Filtered search (only find electronics)
filtered_results = client.query_points(
    collection_name,
    query=query_vector,
    query_filter=models.Filter(
        must=[models.FieldCondition(key="category", match=models.MatchValue(value="electronics"))]
    ),
)
print("Filtered search results:", filtered_results)

**[Day 0] Building Your First Vector Search System**

**High-Level Summary**
- **Domain:** “I built a vector search for [topic]”
- **Key Finding:** “[one sentence on what your vectors captured well]”

**Project-Specific Details**
- **Vector meaning:** d1=…, d2=…, d3=…, d4=…
- **Collection:** day0_first_system (Cosine), points: [count]
- **Query vector:** [a, b, c, d]
- **Top matches (id → score):**
  1) [id] → [score]
  2) [id] → [score]
  3) [id] → [score]
- **Filter used:** category=electronics
- **Filtered result:** [ids returned]

**Why these matched**
- [brief note about direction in 4D space]

**Surprise**
- “[one thing you didn’t expect]”

**Next step**
- “[what you’ll try tomorrow]”